Triple Redundant Power Supply Issue

[bnfdvn]

I was called in to help trouble shoot a control system console that had noise issues. It turned out to be the triple redundant power supplies. Each power supply goes through a diode before being tied together to power seven cabinets. The power supplies are Phoenix Quint 40 Amp supplies and the diodes are rated for 50 Amps. With one power supply powered up, there are no noise issues. As soon as I power up another power supply, the noise issues start. It doesn't matter which power supply is used, the same thing happens when a second is brought up. It is worse with all three power supplies powered up.

Where I saw the noise was on some simulation equipment that is connected to simulate the I/O. The valve commands go to a comparator and the output of the comparator is five volts with frequent noise spikes going to zero volts. I don't see the noise on my scope (200 Mhz). The active signal to the comparator is a 24 VDC PLC output. There is a 10 K Ohm pull-down resistor on the comparator input for when the PLC output is off.

Any suggestions??? I need to get the system to operate with all three power supplies working.

 

[mneary]

With one power supply powered up, there are no noise issues.

This is true with any one power supply?

 

[bnfdvn]

It doesn't matter which power supply is powered up, the noise is not there. As soon as I power up another power supply, the noise is there. It doesn't matter what combination of power supplies are used, if more than one are powered up, the noise is there.

 

[Hero999]

It sounds like the switching frequencies of the power supplies are beating with one another causing other frequencies to be produced.

The problem is, your power supplies were never intended to be used in this way.

Try adding a ferrite bead to each supply and connecting a 1000µF electrolytic and a 100nF ceramic capacitor in parallel with the connection after the diodes.

 

[bnfdvn]

The problem is, your power supplies were never intended to be used in this way.


What do you suggest we use for this application?? Do we need to use a special type of power supply for redundancy?? I don't understand.

 

[Hero999]

Google for switched mode power supply and it might become more clear.

You need some power supplies which support paralleling. You might need to connect them together with a separate data cable for synchronization. The idea is the first one to power up becomes the master and the others are slaves. If the master fails then the other PSUs need some of deciding which one takes over as master. Luckilly you don't have to worry about this, just get some power supplies which support paralleling and they'll sort it out.

 

[MRCecil]

Those 50A diodes, of which you wrote, are no doubt steering diodes to isolate one supply from the others. Try checking all three for reverse leakage and other anomalies. Rectifiers rated at that ampacity are not single junction devices.

 

[bnfdvn]

I found this on the Phoenix Quint power supply website:

"Built-in load sharing capability for parallel operation and increased output current"

Looks to me like these power supplies are designed for use in an application like this.

 

[Hero999]

If that's the case then the diodes aren't needed and could be what's causing the problem in the first place. Unless the manufacturer has suggested you use diodes, I'd recommend removing them, otherwise, take your query to the manufacturer.

 

[MRCecil]

Exactly. That is part of the purpose of the steering diodes; parallel supply operation with increased current when demanded by the load.

Picture two or three supplies bussed together without steering diodes. Now consider a failure of one supply. Depending on the mode of failure of course, the other supply(s) could see another load and this could bring on a cascading failure. The primary purpose of steering diodes, among a few minor reasons, is to prevent cascading failure mode of redundant parallel sources by isolation via the reversed biased diode of the dead supply.

To isolate the offending supply, if it were me, I would physically disconnect one of the three supplies from the buss, light off the remaining two and check for the noise issue. Repeat this until the problem supply is determined by elimination.